A COMPOSITE TRAPEZOIDAL AND SIMPSON RULE BASED TYPE2 FUZZY SYSTEM FOR ENHANCING AFDOA-HETRO-FRBCS
Prediction of diseases at an early stage is the most effective way of increasing the survival rate of people. Various data mining techniques have been proposed for early prediction of disease. One of the most efficient methods for disease prediction is Auto tuned hybridized Firefly and Differential search evolution Optimization Algorithm with Heterogeneous Fuzzy Rule-Based Classification System (AFDOA-Hetro-FRBCS). In FDOA-Hetro-FRBCS, the most representative features in the dataset were selected using a hybridized optimization algorithm and its randomness parameters are fine tuned by AFDOA. The selected features were given as input to Hetro-FRBCS which generates rules for disease prediction. However, the value of membership degree used in FRBCS might include uncertainty. In order to solve the uncertainty problem in AFDOA-Hetro-FRBCS, Type-2 FRBCS (T2FRBCS) is proposed for disease prediction. In T2FRBCS, the value of membership function is given by a fuzzy set that increases the fuzziness of a relation. Hence it has the ability to handle the inexact information in a logically correct manner. One of the processes in T2FRBCS is type reduction (TR) it represents a single value as a representative of the uncertainty. In this paper, composite Trapezoidal rule with Weighted Enhanced Karnik-Mendel (TWEKM) and composite Simpson rule with WEKM (SWEKM) methods are used to perform TR for TFRBCS. It enhanced the defuzzification process in TFRBCS. The whole process is named as AFDOA-Hetro-Type-2 with TWEKM FRBCS (AFDOA-Hetro-T2TFRBCS) and AFDOA-Hetro-Type-2 with SWEKM FRBCS (AFDOA-Hetro-T2SFRBCS) those enhanced the accuracy of disease prediction by solving the uncertainty problem.